Abstract

Organization such as hospitals, publish detailed data or micro data about individuals for research or statistical purposes. Many applications that employ data mining techniques involve mining data that include private and sensitive information about the subjects. When releasing the micro data, it is necessary to prevent the sensitive information of the individuals from being disclosed. Several existing privacy-preserving approaches focus on anonymization techniques such as generalization and bucketization. Recent work has shown that generalization loses considerable amount of information for high dimensional data, the bucketization does not prevent membership disclosure and does not make clear separation between quasi-identifying attributes and sensitive attributes. In this work a novel technique called Fragmentation is proposed for publishing sensitive data with preventing the sensitive information of the individual. Here first the vertical Fragmentation is applied to attributes. In vertical Fragmentation, attributes are segmented into columns. Each column contains a subset of attributes. Secondly, the horizontal Fragmentation is applied to tuples. In this, tuples are segmented into buckets. Each bucket contains a subset of tuples. Finally the real dataset is used for experiments and the results show that this Fragmentation technique preserves better utility while protecting privacy threats and prevents the membership disclosure.